Cyclization of Hydrocarbon Chains Attached to a Planar Chromophore

Cyclization of Hydrocarbon Chains Attached to a Planar Chromophore. Douglas S. Saunders, and Mitchell A. Winnik. Macromolecules , 1978, 11 (1), pp 25â...
2 downloads 0 Views 1MB Size
Vol. 11,No. 1,January-February 1978

Cyclization of Hydrocarbon Chains 25

Cyclization of Hydrocarbon Chains Attached to a Planar Chromophore Douglas S. S a ~ n d e r and s ~ ~Mitchell A. Winnik* Lash Miller Laboratories and Departments of Chemistry, Erindale College and University of Toronto, Toronto, Canada M5S 1 A l . Received June 23, 1977

ABSTRAC!T: A diamond-latticemodel and Monte-Carlo methods are used to simulate cyclization of hydrocarbon chains up to 30 carbons in length attached to a bulky aromatic chromophore. The chromophore is modeled in the lattice by specifying as occupied those sites which best resemble its geometry. Cyclization is detected a t a site adjacent to the chromophore (called the “reactive volume”) somewhat remote from where the chain is attached. Furthermore, a chain which has any element in the reactive volume is deemed as cyclized. These choices simulate intramolecular photohydrogen abstraction reactions in n-alkyl esters of benzophenone 4-carboxylate.Partition functions are calculated separately for both cyclized and all chains. Their ratio, the cyclization probability P,, is estimated a t several temperatures for real hydrocarbon chains ( E g= Et t 500 cal/mol) and for hypothetical chains biased to favor gauche rotamers (Eg= Et and E , = Et - 500 cal/mol). For the system under consideration, cyclization probability is relatively insensitive to extremes of temperature and to the statistical weights given gauche rotamers in the chain. When a hydrocarbon chain is attached to a large bulky chromophore certain regions of space are no longer accessible to the chain. The substituent excludes a volume of space that would be available tlo an unsubstituted chain, and this steric effect must manifest itself in the conformational properties of the chain. One might imagine that various conformational properties of the chain might be differently sensitive to the presence of the chromophores. One might speculate that chain properties sensitive primarily to more extended chain configurations, like the mean-squared end-to-end distance1 ( r2) and higher even moments of r, might be relatively insensitive to the chromophore. On the other hand, chain properties more sensitive to compact configurations (cyclization, the persistence vector2 a, and inverse moments3 of r) might be much more sensitive both to the presence and the nature of the terminal substituent. Cyclization probability4 should be the conformational property of flexible (chainsmost sensitive to interactions between remote chain elements and to substituents attached to the chain. There is chemical evidence from the early work of Carothers, Ziegler, and others on medium size ring formation (10 to 16 carbons) that transannular interactions between CH2 groups affect cyclizakion of the chain terminie5On the other hand, the presence of oxygens or sp2 carbons in the chain mitigate some of these remote interactions, and the odd-even alternating effect in those particular cyclization reactions disappears. We are interested in determining what factors affect the cyclization of hydrocarbon chains. In this paper, we explore aspects of cyclization of hydrocarbon chains substituted at one end by a bulky planar chromophore. The approach taken is that of conformational calculations based upon a model which allows the bulk of the terminal substituent to be accomodated and which also excludes chain configurations in which one element of the chain would interfere sterilcally with a second element of the chain. Thus, within the context of the model, excluded volume considerations are taken into account. reactive volume

0024-9297/78/2211-0025$01.00/0

We conceptualize the problem in I. The bulky substituent is indicated, and C1 serves as a convenient origin for our coordinate system. Since, experimentally, cyclization is detected by chemical reactions, we call the volume (labeled F) in which one would detect cyclization the “reactive volume”. In a simple polymer not appended to a substituent, the reactive volume would be adjacent to C1. One can consider three kinds of cyclization probability for chains attached to a substituent. (i) One could focus on the probability that the chain terminus, the methyl group n , occupies the volume of space designated to imply cyclization (F in I); (ii) one could consider the probability P,,, that the ith CH2 group in an n-carbon chain occupied that volume; or (iii) one could consider the sum of probabilities P, = Z,P,,, that any chain element occupied the volume F . In this paper, we describe the effects of chain length, temperature, and the gauche-trans rotational energy difference on those kinds of cyclization probability described by (ii) and (iii). Our choice is based upon the availability of experimental and P,. These form a basis evidence closely related to PL,, from which predictions of our model may be evaluated. The model involves random walks on a tetrahedral lattice with second neighbor exclusions. Monte-Carlo techniques coupled with the sampling method of Rosenbluth and Rosenbluth7are used to compute averaged properties of the chains. The terminal substituent is accomodated by assigning as occupied those lattice sites which best describe its geometry. Certain lattice sites are chosen to simulate the reactive volume F in I. Chains passing through this volume are counted as cyclized chains. Properties of cyclized chains are compared with those of all chains. Values of Pi,,and P, are obtained from the appropriate estimated partition functions. Similar models have been used by Smiths and by Bothorelg to calculate various properties of the n-alkanes. Smith used exact enumeration methods to evaluate selected properties of hydrocarbon chains up to 18 carbons and Monte-Carlo techniques to examine longer chains.s Several authors1°-14 have used off-lattice methods to explore properties of these chains. Sisido focused on cyclization probability for chains up to 16 carbons and demonstrated the importance of taking account of excluded volume in evaluating cyclization probabilities. With the exception of the very sophisticated model of La1 and Spencer,”* these calculations considered threefold rotational states about carbon-carbon bonds. The degree to which predictions of such models differ for such short chains from predictions based upon diamond lattice models is not known. The major difference between the models is the distortion of the CCC bond angle from 112 to 109.5’ on the lattice.

0 1978 American Chemical Society

26 Saunders, Winnik reactive volume a

Macromolecules

: c : l

Figure 1. Diamond-lattice model of benzophenone 4-CO2(CH&-lCHs showing the first three carbons of the n-alkyl ester. The

benzophenone rings and the ester group are coplanar. The two rotamers of the ester group are designated syn or anti with respect to the keto oxygen, and the ester is in the trans configuration. Equal numbers of chains are grown from each rotamer. The first three carindicate posbons in the chain are labeled C1, CZ,Cg. The circles (0) sible loci for Cq. The sites designated (0)represent the three inner hit sites; those designated ( 0 )represent the seven outer hit sites. If any of these sites are occupied by an H from a CH2 group, that chain is counted as in a reactive (hitting) conformation.

2, ( x + y ) = ( n - 2 )

0 hu

1 (n=l)+

d

b

4

0

C

H

3

k + n-hexane i

COzCH3

+

R.

3, R = C6Hll The molecules we consider are the n-alkyl esters of benzophenon 4-carboxylate, l. The intramolecular photochemistry of these molecules has been studied in some detail in our lab ~ r a t o r y . ~Rate ~ J ~constants k , , were obtained with high precision for the intramolecular hydrogen abstraction reaction 1 2, a process involving cyclization of the chain to occupy the reactive volume in 1*.P, cannot be obtained directly from these results. Rather, we compared values of k , , with that of k2, the rate constant for the corresponding bimolecular hydrogen abstraction reaction. The ratio (kr,,/kz = Ceff), the effective concentration,16has units of molarity. I t is proportional to P, and can be related to P, if reasonable assumptions are made about the size of the reactive volume in l*.When these assumptions are made, the similarity between experimental values of C,ff and those calculated from P, is quite good.l5 Thus the model has had some success a t simulating experimental phenomena.

-.

Methodology and Semiempirical Considerations (a) Structural Parameters. T o describe the hydrocarbon chains, we examined self-avoiding walks on a diamond lattice with second-neighbor exclusions. Some structural parameters (the bond angle 0 between successive CH2 units and the rotational angle 4) are dictated by the lattice. Others were introduced empirically: the C-C bond length 1 and the energy difference EG = E , - Etfor gauche and trans rotamers of the chain. Successive gauche bonds of the opposite sense are prohibited by the second-neighbor exclusion feature of the random walks. Following Flory,2 we used a value of 500 cal/ mol for EG. The ester group in the substituent is known to be planar. Steric and dipolar effects keep it in the trans configuration., Thus, only that configuration is considered in this work (see Figure 1).The asymmetry of the benzoyl group in the substituent gives rise to two rotamers about the ester group of virtually identical energy. These place the chain bearing oxygen either syn or anti to the ketone carbonyl in the substituent. We generate equal numbers of chains from the syn and anti ester groups and weight them equally in all calculations.

The addition of a bulky substituent to one end of the chain changes the rotational energy potentials of the first three methylene groups from those of an unsubstituted polymethylene chain. The ester group plays the major role in these changes. Flory has examined the relative energies of the rotamers about the 0-CH2 bond in his work on poly(ethy1ene terephthalate).18 Gauche rotamers place the second methylene group 2.83 8, from the carbonyl oxygen, within the sum of their respective van der Waals radii ( r m 2 ro) = 1.7 1.5 = 3.2 &. Flory calculates that there are 400 cal of extra energy, relative to the trans orientation, associated with these rotamers. The uncertainty of the calculation, and the convenience of treating this interaction in a manner similar to gauche methylenes within unsubstituted polymethylene chains, has led us to assign EG = 500 cal/mol to these gauche rotamers. The interactions of a methylene group and an ether oxygen, separated by three bonds, have been examined by Mark in his studies of poly(ethy1ene oxide) and poly(tetramethy1ene o ~ i d e ) In . ~a~detailed , ~ ~ analysis of these chains, he assigned a stabilization energy to the gauche rotamers of 200 cal/mol, relative to a methylene group oriented trans. He suggests that this is a result of small coulombic interactions between the methylene group and the oxygen, separated by 2.9 A. Steric repulsions, he postulates, are unimportant a t this distance. Comparison of his theoretical work with experiment suggests that gauche and trans rotamers could be isoenergetic (Le., E , = -200 f 200 cal). Since, experimentally, polyoxyalkanes are not as well understood as polymethylene chains, the present work assigns gauche and trans states of this step equal statistical weights. Certain conformations about bonds (i), (i l),and (i + 2) place C(4)Hz of the chain rather close to the ester oxygen to which C(l)H2 of the chain is attached. In these conformations, the bonds (i 1)and (i 2) are g*gF. Mark has analyzed an identical geometry in poly(tetramethy1ene oxide).lgHe suggested that these conformations are destabilized by an additional 340 f 250 cal/mol. Because of the large uncertainty in this value,20we choose for simplicity to assign 500 cal/mol to each gauche conformation in the sequence. (b) The Chains and the Substituent. A diamond lattice is not an ideal framework for simulating the bulk of a planar chromophore; nevertheless, by designating as occupied those vertices which best correspond to the shape of the chromophore, one can map out a region of space similar in shape to that of the substituent under consideration. In this work, benzophenone 4-carboxylate could be simulated by dicyclohexyl ketone 4-carboxylate, in which only the equitorial hydrogens were kept in the lattice model, Figure 1.Space-filling models show a remarkable similarity in the shapes of the

+

+

+

+

+

C V C ~ ~ Z PO?~ Fy?rocarbon ~CI;, Chains 27

Vol. 11, No. 1,January-February 1978

Cyclization Probability P,

12

14 16 18

20 22

24 26 28 30

4.89 (h0.23) 10.17 (h0.41) 12.87 (30.55) 15.99 (40.85) 20.02 (31.13) 23.11 (31.42) 26.14 (Q1.81) 33.45 (43.87) 38.73 (h6.03) 37.37 (*7.05)

4.50 (f0.20) 9.66 (f0.33) 13.00 (f0.43) 16.60 (60.63) 20.76 (f0.86) 24.04 (f1.06) 27.00 (f1.34) 32.66 (52.58) 36.58 (f3.62) 36.21 (54.29)

Table I ( j ) vs. Chain Length" (Inner Hits) - ___ __ -..__ _- -- - -

1.82 (+0.01) 6.17 (f0.10) 10.66 (f0.13) 15.76 (f0.17) 20.73 (50.21) 25.20 (f0.25) 28.98 (10.28) 32.35 (f0.30) 35.29 (f0.33) 35.98 (f0.35)

0 04 0.97 (f0.07)

0.39 (40.01) 2.86 (f0.06) 6.17 (f0.11) 11 26 (d.0.18) 16.70 (f0.26) 22.30 ( f 0 35) 26.21 (f0.41) 30.09 (10.40) 33.89 (f0.47) 34.56 (10.49)

2.41 (f0.13) 6.31 (f0.39) 11.04 (f0.75) 21.05 (41.52) 20.26 (k1.75) 20.41 (f1.82) 22.62 (f1.89) 21.49 (k1.95)

' Values X104;parentheses contain error estimates, reported as 2a. Values repoited x e : Z (kilttirig Lhalns)/l rail h i i n s ) ; AE = 0.5 or -0.5 kcal/mol. ' For E t = E,, values reported are also equal to (estimated number of hitting cbsinq)/iestimsted number of all chains). Table I1 Cyclization Probability P,, ( O ) vs. Chain Length" (Outer Hits)

__

~

Chain length 11

13 15 17 19 21 23 25 21

29

Et

< E,,b T

-20°C

1.41 (f0.05) 2.93 (10.10) 4.17 (f0.15) 5.08 (f0.18) 6.01 (34.25) 6.60 (f0.30) 7.56 (f0.55) 8.41 (2~0.55) 9.09 (fO.70) 9.32 (f0.98)

=

25 "C 1.44 (f0.05) 3.05 (f0.08) 4.41 (f0.12) 5.50 (50.16) 6.47 (f0.20) 7.17 (50.24) 8.07 (f0.36) 8.93 (10.40) 9.53 (f0.52) 9.96 (50.78)

E t = E,'

1.34 (f0.03) 2.93 (f0.04) 4.59 (10.05) 6.27 (10.06) 7.67 (50.06) 8.92 (f0.08) 9.95 (f0.08) 10.86 (10.08) 11.66 (f0.09) 12.37 (fO.10)

-~

25 "C

~

E, :Er.O 7' ----$5 "7

1.03 (f0.02) 2.09 (f0.03) 3.61 (f0.04)

5.34 (33.06) 7.03 (10.07) 6.57 (f0.10) 9.84 ( f O . l O ) 10.95 (f0.11) 11.88 (f0.13) 12.79 (40.14)

-128 __

"c ___

0 79 (f0.03) 1.25 (50.05) 2 35 (*0.08) 3.96 (10.18) 6.12 (40.36) 7.62 (f0.48) 8.36 (f0.50) 8.24 (f0.52) 9 76 (20.64) 10,5'? (f0.80)

a Values X103; parentheses contain error estimates, reported as 20. Values reported are: 2 (hitting chains)/Z (all chains); 3E = 0.5 or -0.5 kcal/mol. For E t = E,, values reported are also equal to (estimated number of hitting chsirlsl/iestirnated number of all chains).

chromophore and its modeLpl Chains were grown stepwise from the ester oxygen as indicated in Figure 1. C1 was placed in the trans geometry with respect to the ester. 122 was placed a t random in one of the three lattice sites a d j x e n t to C1. The remaining valences on C1 were then designated as occupied by hydrogens. C3 was then added a t random to one of the lattice sites adjacent to Cp; then its valences were completed with the addition of hydrogens. In this manner, chains were grown up to 30 carbons.21 T o sample all chains attached to the substituent, two samples of 180 OOO chains were grown. One sample was grown from the syn conformation of the ester and the other was grown from the anti conformation. In addition, separate samples of nearly 400 000 chains were grown with a pseudopotentional importance sampling techniquez1to generate 60 000 cyclized chains from each conformer of the ester. Chains were grown with the probability of choosing a gauche step, p,, equal to pt, that of choosing a trans step. Under these conditions, E , = Et. The coordinates of each carbon and its step weight were stored on magnetic tape. Real temperatures were introduced into the calculations by reweighting gauche vectors by a factor of exp(-500/RT). Monte-Carlo methods were used to estimate properties of the chains. Details of the calculation of the standard deviation in P, are described in the Appendix. ( c ) Detecting Cyclization. Figure 1 shows the lattice representation of the substituent from which samples of chains were grown. Cyclization was detected by identifying those chains which hald an H from a CHp group occupy one of the lattice sites designated as reactive. We call those sites "hit sites" and the chains which pass through them "hitting

chains". Ten such sites are indicated in Figure 1.. These surround the ketone carbonyl oxygen. Because only even-membered rings may be formed on a tetxahedral lattice, we could count separately those chains that occupy one of the three hit sites adjacent to the reactive oxygen from those that occupy the outer tier of seven hit sizcjs. By comparing the partition functions of cha.ins passing through the inner hit sites with those passing through the latter, Outer, hit sites, we get some indication of the sensitivity of cyclization t u the size of the volume designated 8 5 reactive. Results and Discussion

In the photochemical reactions of i, hydrogen abstraction can be detected ondy at t h e CH, groups iri the chain; the methyl groups are negligibly reactive. Therefore, we define the cumulative cyclizatioii probability P, a.s the sum of cyfor each of the ( n - " 1) CIIp groups clization probabilities P,,,, in the chain. We can (;unbid~r:rtwo classes o f cyclized chains. For those that have an 13 from a. (?:& group occupying one of the three lattice sites adjacent t u the reactive oxygen hit a t the "inner hit sites", we (denotethe cyc!ization probability of these chains as P,(i). For those that have a methylene hydrogen occupying one of the adjacent tier of seven lattice sites, the "outer hit sites", we denrjte the cyclization probability Pn("). When we wish to consider- the partition function for the separate classes of chains which u p y the inner or outer hit sites, we shall use the superscrip+,forms Zo(i)and Z,(O),respectively. The cyclization probahility is equal to the ratio of partition functions for cyclized chains and dl chains. In the special case that E , = E,, the partitior, furletions become simply the

28 Saunders, Winnik I

1

0

d

Macromolecules

Et

'

1

200

0 -

1

I Io

f

.ot 3

II 15 19 23 CHAIN LENGTH

7

c

27

Figure 3. The chain length dependence of cyclization probability at the outer hit sites for ( A ) E t = E,; (0) E , = Et t 500 cal/mol, 25 "C, and (0) -20 "C.

P,(O)

CHAIN

LENGTH

Figure 2. The chain length dependence of the cyclization probability P,(i)at the inner hit sites for n-alkyl esters of benzophenone 4-carE , = Et 500 cal/mol, 25 "C; (0) E , = Et boxylate: ( A ) Et = E,; (0) - 500 cal/mol, 25 "C. The error bars represent two standard devia-

1

4

.

+

0

"

~

'

~

T=25'

Eg